Causal inference specialists for your dedicated team.

Where causal inference fits in a long-term product team.

A great deal of what businesses want to evaluate cannot be randomised. Pricing changes, market launches, regulatory effects and brand campaigns are rolled out to everyone at once, and the question of what they caused still deserves an answer. Quasi-experimental methods are how that is done credibly.

Every method rests on assumptions the data alone can't test, such as parallel trends, a valid instrument or no unmeasured confounding. Good practice is stating those assumptions plainly and testing what can be tested. An estimate presented without its assumptions is an opinion with a confidence interval attached.

What your dedicated team does with causal inference.

The failure mode is not the arithmetic; it is applying a method whose assumptions the situation violates. That judgement comes from doing the analysis in context many times, well beyond knowing the technique.

It also depends on knowing the business well enough to name the plausible confounders, which is product knowledge accumulated over time. What that continuity commitment involves is set out in how the monthly fee is built.

Common causal inference use cases.

• Evaluating a change that was not split Difference-in-differences or synthetic control where a rollout could not be randomised.
• Assumptions stated with the estimate What the result depends on, made explicit.
• Confounders identified from domain knowledge Knowing what else changed, which no method can discover on its own.

Adding causal inference skills to your team.

Causal analysis sits with a dedicated data science team, with the method's assumptions reported alongside every estimate.

You can add causal inference capacity as one dedicated specialist or as part of a wider remote team, managed by a service delivery manager from our offices in Thailand. Companies hire dedicated developers this way under one agreement and one monthly fee.